G violence and mental illness clinical vs statistical

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: none—particularly problematic when base rates are low—e.g., violence and mental illness Ø Clinical vs. Statistical Prediction Judgments and Formulas Are the predictions of human experts more reliable than the predictions of actuarial models? Clinical vs. Statistical Prediction ◦ Statistical (actuarial/ nomothetic) method - classifying the individual among similar cases ◦ Clinical (idiographic) method - attempting to understand the individual apart from other cases The Clinical Method Advocates of the clinical method: ◦ ◦ ◦ Argue that the only way to make a certain prediction of cause and effect is to study the life in which the causes operate, and not a thousand other lives. Psychological causation is always personal and never actuarial. Gordon Allport (1942, p.157) The use of personal documents in psychological science The Clinical Method Allport's example: If we know that 80 percent of delinquents from broken homes are recidivists, it does not follow that a single delinquent from a broken home has an 80 per cent probability of becoming a recidivist. Why? ◦ ◦ Probability is based on the frequentist notion of repeated trials. The delinquent here represents only a single trial! This individual has a 100% or 0% chance of becoming a recidivist. However, if we know all the variables that are related to recidivism then we would be able to predict the delinquent's outcome with certainty. The Statistical Method Paul Meehl's "Disturbing little book" , Clinical Versus Statistical Prediction: A Theoretical Analysis and Review of the Literature (1954). Ø According to Allport, we should discard all predictions based on class membership, e.g., actuarial predictions. He argued that only if all predictors were known, it is OK to use predictors, e.g., class membership, to predict recidivism. For then, we would predict with reasonable (near 100%) certainty. Ø Meehl disagreed with this. Ø Why not predict (using regressions) with 80% certainty? 70% certainty? etc.... The Statistical Method The most disturbing chapter in Meehl's disturbing little book: ◦ Review of empirical comparisons of statistical and clini...
View Full Document

This note was uploaded on 01/31/2014 for the course PSYC 433 taught by Professor Dr.k.dobson during the Winter '13 term at University of Calgary.

Ask a homework question - tutors are online